Press Office

Fake news

Filtering out fake news

Published on: 7 March 2018

Why the need for evidence-based research is more important than ever before.

Every second we are bombarded with claims and counter-claims on every possible topic from climate change to cancer treatments.

Through the proliferation of social media, everyone has an opinion and many of us are reading, sharing and believing information that is factually inaccurate and in some cases purposely fake.

When some scientific findings report one result, and others the opposite, this can leave us at best confused and at worst reaching the wrong conclusions and making bad decisions. This can affect our health, our environment, and our society, with sometimes very poor outcomes.

Risk making the wrong decisions

“Claims with no scientific proof cast doubt over those with overwhelming evidence, leaving us at the best confused and in the worst case making totally the wrong decisions - particularly when the information has an impact on policy,” explains Dr Gavin Stewart, an expert in evidence synthesis from Newcastle University, UK.

“Climate change is a case in point.  There is now overwhelming scientific evidence that climate change is real and already having a major impact on the world.

“But because we are so keen to be balanced and hear both sides of the story, we give credibility to the contradictory arguments.  So, despite there being no scientific evidence to back up the claims of the climate sceptics, they cast doubt on the truth.”

Publishing a paper on the subject today in Nature, Dr Stewart, from Newcastle University, and co-authors Professor Julia Koricheva, from Royal Holloway University London, Professor Jessica Gurevitch from Stony Brook University (USA) and Dr Shinichi Nakagawa from the University of New South Wales (Australia) champion the use of meta-analysis – the backbone of evidence-based research.

“The findings of a single scientific study may seem to tell us that a particular phenomenon or effect exists”, explains Prof Koricheva.

“But any individual study is subject to potential limitations and biases. Particular outcomes may result by chance, especially when the number of people or samples studied are small, and various biases may tip the conclusions. Fortunately, over the last few decades statistical methods have been developed to more accurately and objectively summarize the results of multiple studies on the same topic.” 

Forty years of meta-analysis

Meta-analysis is the quantitative, scientific synthesis of research results – the mathematical system which underpins evidence-based research. First introduced in the 1970s, it has had a revolutionary impact on many scientific fields, helped to establish evidence-based practice and resolve seemingly contradictory results.

Used as a way to reach a broad understanding of a general problem – building an accurate picture by analysing dozens to thousands of related studies – meta-analysis has been a more powerful and less biased means of confirming – or disproving – assumed wisdom.

Marking forty years of meta-analysis, Dr Stewart and colleagues say that in today’s age of social media and the ‘fake news furore’, we need to be using it to drive not just medicine, but all policy and research.

“Instead of relying on ground-breaking, individual studies, it allows us to take apparently inconclusive data and arrive at a clearer picture,” explains Dr Stewart.

“Since MMR, meta-analysis has become embedded in medical research and is the basis of robust decisions that are made around drug effectiveness and new treatments.

“But we need to be using it across the board to solve the big challenges facing society, such as climate change, education and future food security.

“If we can underpin these decisions with trustworthy data, looking at the evidence from hundreds of studies not just one spurious claim, then we can start to counter the fake news.”


'Meta-analysis and the science of research synthesis'. Jessica Gurevitch, Julia Koricheva, Shinichi Nakagawa and Gavin Stewart. Nature


Latest News